Intelligent optimization method and system for metal additive manufacturing process parameters based on Bayesian optimization

Abstract

The invention provides an intelligent optimization method and system for metal additive manufacturing process parameters based on Bayesian optimization, and the method comprises the steps: testing the performance parameters of samples prepared under different additive manufacturing process parameters, and constructing a data set; based on the data set, training a single performance prediction model through a Gaussian regression algorithm, performing weighting processing on a plurality of performance parameters to obtain comprehensive performance, and training a comprehensive performance prediction model; performing Bayesian optimization by taking the comprehensive performance prediction model as an agent model; determining recommended process parameters according to a Bayesian optimization result, and predicting a predicted value of the sample under the recommended process parameters through a single performance prediction model; and based on the error between the measured value and the predicted value of the prepared sample, updating the data set and circularly carrying out model training and Bayesian optimization until the process parameters meeting the requirements are obtained. According to the method, a complex additive manufacturing process parameter space can be efficiently explored, and a plurality of conflicting additive manufacturing quality indexes are collaboratively optimized.

Description

Technical Field

[0001] This application relates to the field of metal additive manufacturing technology, and in particular to a method and system for intelligent optimization of metal additive manufacturing process parameters based on Bayesian optimization. Background Technology

[0002] Currently, additive manufacturing, especially metal additive manufacturing, as an advanced manufacturing technology, has been widely applied in aerospace, biomedicine, industrial mold making, and power energy fields. Among these, the quality of the final products obtained from metal additive manufacturing processes such as laser powder bed melting and laser direct energy deposition is affected by various process parameters, and these parameters exhibit non-linear coupling relationships, limiting the development efficiency of metal additive manufacturing processes and the improvement of metal component quality. Therefore, it is necessary to optimize the process parameters of metal additive manufacturing.

[0003] In related technologies, optimization methods for metal additive manufacturing process parameters mainly rely on experience, trial and error, or limited experimental designs (such as orthogonal experiments). In practical applications, these optimization methods suffer from the following problems: First, high cost and low efficiency: each process experiment is time-consuming and costly, resulting in high trial and error costs and making it difficult to meet the requirements of rapid product iteration. Second, difficulty in coordinating multiple objectives: because optimization methods in related technologies struggle to automatically find the optimal trade-off point, various optimization objectives often conflict, leading to optimization results that fail to meet actual production needs. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] Therefore, the first objective of this application is to propose an intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization. This method can significantly reduce the number of expensive experiments in the optimization process of metal additive manufacturing process parameters, effectively resolve multi-objective conflicts, improve additive manufacturing quality and process development efficiency, and can be widely applied to metal additive manufacturing processes such as laser powder bed melting and directional energy deposition, thus solving the high cost problem of multi-objective parameter optimization in metal additive manufacturing processes.

[0006] The second objective of this application is to propose an intelligent optimization system for metal additive manufacturing process parameters based on Bayesian optimization.

[0007] The third objective of this application is to propose an electronic device.

[0008] The fourth objective of this application is to provide a computer-readable storage medium.

[0009] To achieve the above objectives, the first aspect of this application is to propose an intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization, comprising the following steps:

[0010] Using the additive manufacturing process to be optimized, multiple sets of metal samples were prepared under multiple process parameters with different values, and multiple performance parameters of the multiple sets of metal samples were tested to construct a dataset reflecting the relationship between process parameters and performance.

[0011] The dataset is used as training data, and the multiple process parameters are used as model inputs. A single performance prediction model for predicting individual performance parameters is trained using a Gaussian regression algorithm. The multiple performance parameters are then weighted to obtain the comprehensive performance. A comprehensive performance prediction model for predicting the comprehensive performance is then trained using a Gaussian regression algorithm.

[0012] Set the range of the multiple process parameters, use the comprehensive performance prediction model as a surrogate model in Bayesian optimization, select a data acquisition function for the additive manufacturing process, and perform Bayesian optimization based on the range and the data acquisition function.

[0013] Based on the Bayesian optimization results, the recommended process parameters that make the objective function of the comprehensive performance prediction model optimal are determined, and the predicted values ​​of each performance parameter of the metal sample under the recommended process parameters are predicted by the single performance prediction model.

[0014] The performance of the additive manufacturing samples prepared under the recommended process parameters is measured. Based on the error between the measured value and the predicted value, the dataset is updated and the prediction model is trained and Bayesian optimization is performed iteratively until the corresponding process parameters that meet the error requirements are obtained.

[0015] Optionally, the additive manufacturing process includes: laser powder bed melting, electron beam powder bed melting, and direct energy deposition, and the multiple performance parameters include: density, hardness, yield strength, tensile strength, elongation, surface roughness, and residual stress.

[0016] Optionally, when the additive manufacturing process to be optimized is laser powder bed melting, the multiple process parameters include: laser power, scanning speed, powder layer thickness, scanning spacing, powder particle size, and substrate preheating temperature. Constructing a dataset reflecting the relationship between process parameters and performance includes: obtaining samples from the multiple groups of metal samples; conducting corresponding tests on the samples for the multiple performance parameters to be tested; obtaining the values ​​of multiple performance parameters for the metal samples prepared under each group of process parameters; summarizing the values ​​of each group of process parameters and the corresponding multiple performance parameters to generate an additive manufacturing process parameter and performance parameter data table; and normalizing the data table to obtain the dataset reflecting the relationship between process parameters and performance.

[0017] Optionally, the step of weighting the multiple performance parameters to obtain the comprehensive performance includes: setting a weighting coefficient for each performance parameter based on user preferences and the importance of each performance parameter; multiplying each weighting coefficient by the value of the corresponding performance parameter; accumulating the obtained multiplications to calculate the comprehensive performance index; and using the comprehensive performance index as training data for the comprehensive performance prediction model.

[0018] Optionally, the acquisition function includes: expected improvement, probabilistic improvement, and upper confidence limit. When the additive manufacturing process to be optimized is laser powder bed melting, the acquisition function selected is the expected improvement.

[0019] Optionally, updating the dataset based on the error between the measured value and the predicted value and iteratively training and optimizing the prediction model includes: taking the current recommended process parameter as the final process parameter when the error corresponding to each performance parameter is within a preset allowable range; adding the measured value and the recommended process parameter to the dataset to update the dataset when the error corresponding to any performance parameter exceeds the allowable range; retraining the single performance prediction model and the comprehensive performance prediction model using the updated dataset, and performing Bayesian optimization again on the retrained comprehensive performance prediction model.

[0020] Optionally, the error is the ratio of the absolute value of the difference between the measured value and the predicted value to the predicted value; the allowable range is less than or equal to 10%.

[0021] To achieve the above objectives, a second aspect of this application also proposes an intelligent optimization system for metal additive manufacturing process parameters based on Bayesian optimization, comprising the following modules:

[0022] The module is used to prepare multiple sets of metal samples under multiple process parameters with different values ​​using the additive manufacturing process to be optimized, and to test multiple performance parameters of the multiple sets of metal samples to construct a dataset reflecting the relationship between process parameters and performance.

[0023] The training module is used to use the dataset as training data, the multiple process parameters as model inputs, train a single performance prediction model for predicting a single performance parameter using a Gaussian regression algorithm, and perform weighted processing on the multiple performance parameters to obtain the comprehensive performance, and train a comprehensive performance prediction model for predicting the comprehensive performance using a Gaussian regression algorithm.

[0024] The optimization module is used to set the range of the multiple process parameters, use the comprehensive performance prediction model as a surrogate model in Bayesian optimization, select a data acquisition function for the additive manufacturing process, and perform Bayesian optimization based on the range and the data acquisition function.

[0025] The prediction module is used to determine the recommended process parameters that make the objective function of the comprehensive performance prediction model optimal based on the Bayesian optimization results, and to predict the predicted values ​​of each performance parameter of the metal sample under the recommended process parameters through the single performance prediction model.

[0026] The update module is used to perform performance tests on the additive manufacturing samples prepared under the recommended process parameters, update the dataset based on the error between the measured values ​​and the predicted values, and iteratively train and optimize the prediction model until the corresponding process parameters that meet the error requirements are obtained.

[0027] To achieve the above objectives, a third aspect of this application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the Bayesian-based intelligent optimization method for metal additive manufacturing process parameters as described in any one of the first aspects above.

[0028] To achieve the above objectives, the fourth aspect of this application also proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization as described in the first aspect above.

[0029] The technical solution provided by the embodiments of this application brings at least the following beneficial effects: This application uses a Gaussian process regression model to model multiple objectives separately, and uses normalization and user preference weighting to transform complex multiple objectives into a single objective; based on this single objective, a Bayesian optimizer is driven to efficiently recommend the optimal combination of process parameters; the process parameter optimization model is automatically iterated according to the deviation between the measured and predicted performance values ​​under the optimal combination of process parameters, improving the quality of the final recommended process parameters and the credibility of the decision. Therefore, this application addresses the problems of high cost, long cycle, and small dataset in acquiring additive manufacturing process parameter-performance data by using Gaussian process regression modeling, reducing the number of experiments and the amount of data samples to be collected, thus reducing the cost and time of process parameter optimization and improving process optimization efficiency. Furthermore, for multi-objective optimization problems, this application transforms the multi-objective optimization problem into a single-objective optimization problem, and uses Bayesian optimization to efficiently explore the parameter space, quickly finding the optimal trade-off point between multiple conflicting objectives, and can recommend the optimal combination of process parameters. Furthermore, this application possesses a closed-loop feedback mechanism from prediction to verification, which can automatically iteratively optimize the process parameter optimization model based on the deviation between the measured value and the predicted value, thereby improving the quality of the final recommended process parameters and the credibility of the decision.

[0030] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0031] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0032] Figure 1 This is a flowchart of an intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization proposed in an embodiment of this application;

[0033] Figure 2 A flowchart illustrating a method for constructing a dataset according to an embodiment of this application;

[0034] Figure 3 This is a flowchart illustrating a specific intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization proposed in an embodiment of this application.

[0035] Figure 4 This is a schematic diagram of a process parameter recommendation and performance prediction display interface in an additive manufacturing process parameter optimization system proposed in an embodiment of this application;

[0036] Figure 5 This is a schematic diagram of the verification feedback interface in an additive manufacturing process parameter optimization system proposed in an embodiment of this application;

[0037] Figure 6 This is a schematic diagram of the structure of an intelligent optimization system for metal additive manufacturing process parameters based on Bayesian optimization, as proposed in an embodiment of this application. Detailed Implementation

[0038] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0039] It should be noted that with the widespread application of machine learning technology, machine learning has also made significant progress in accelerating material design and optimizing processing technology. However, most of the methods for optimizing process parameters based on machine learning technology in the relevant embodiments require a large amount of process-performance data, and multi-objective optimization methods based on Pareto fronts converge slowly in high-dimensional spaces and the results are difficult to interpret.

[0040] To this end, this application proposes a Bayesian optimization-based intelligent optimization method for metal additive manufacturing process parameters. This method is an efficient and low-cost intelligent optimization method for additive manufacturing process parameters with small sample data. It can effectively solve the multi-objective conflict problem, reduce the number of process parameter trial and error and multi-objective performance optimization iterations, accelerate the efficiency of additive manufacturing process development, and improve the quality of additive manufacturing products.

[0041] The following description, with reference to the accompanying drawings, illustrates an intelligent optimization method and system for metal additive manufacturing process parameters based on Bayesian optimization, as proposed in an embodiment of this application.

[0042] Figure 1 This is a flowchart illustrating an intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization, as proposed in an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0043] Step S101: Using the additive manufacturing process to be optimized, multiple sets of metal samples are prepared under multiple process parameters with different values, and multiple performance parameters of the multiple sets of metal samples are tested to construct a dataset reflecting the relationship between process parameters and performance.

[0044] Specifically, this application can optimize parameters for different additive manufacturing processes, the specific parameters of which can be determined according to actual needs. Based on the currently optimized additive manufacturing process, the process parameters used when preparing metal samples are determined. This application also obtains data on metal additive manufacturing process parameters of different values ​​and corresponding product performance parameters through actual testing, thereby constructing a dataset to provide data support for subsequent model training.

[0045] That is, firstly, metal samples are prepared using the current optimized additive manufacturing method under multiple sets of process parameters with different parameter values. Then, the performance of metal samples prepared with different sets of additive manufacturing process parameters is tested, and a dataset of metal additive manufacturing process parameters and performance is constructed.

[0046] In one embodiment of this application, the additive manufacturing process includes: laser powder bed melting, electron beam powder bed melting, and direct energy deposition. That is, the additive manufacturing processes for which parameter optimization is possible in this application embodiment include, but are not limited to, the various processes described above. Furthermore, in this embodiment, when the additive manufacturing process to be optimized is laser powder bed melting, multiple process parameters include: laser power, scanning speed, powder layer thickness, scanning spacing, powder particle size, and substrate preheating temperature. That is, this application embodiment can select the various process parameters required to prepare metal samples based on the currently optimized additive manufacturing process. When using laser powder bed melting, the selectable process parameters include, but are not limited to, the multiple process parameters described above.

[0047] In one embodiment of this application, multiple performance parameters of various metal samples are tested, including: density, hardness, yield strength, tensile strength, elongation, surface roughness, and residual stress. That is, this embodiment allows for the selection of performance parameters for each metal sample required to optimize the current additive manufacturing process. For example, when the additive manufacturing process to be optimized is laser powder bed melting, the selectable performance parameters include, but are not limited to, any combination of two or more of the above-mentioned performance parameters. For instance, the multiple performance parameters to be tested for various metal samples may include tensile strength and elongation.

[0048] To more clearly and intuitively illustrate the specific implementation process of optimizing the metal additive manufacturing process parameters in this application, the following is a specific example of optimizing the process parameters in the preparation of aerospace titanium alloy Ti-6Al-4V by laser powder bed melting (LPBF) process.

[0049] In this embodiment, the laser powder bed melting process parameters were first adjusted to prepare a series of Ti-6Al-4V titanium alloy samples. In specific implementation, the process parameters used included laser power, scanning speed, powder layer thickness, powder particle size, overlap rate, and substrate preheating temperature (hereinafter referred to as substrate temperature).

[0050] Then, the performance parameters of multiple prepared Ti-6Al-4V titanium alloy samples were tested, and the data were normalized to construct a laser powder bed melting process parameter-performance dataset for aerospace titanium alloy Ti-6Al-4V.

[0051] To more clearly illustrate the specific implementation process of constructing a dataset reflecting the relationship between process parameters and performance in this application, the following example illustrates a dataset construction method proposed in this embodiment. Figure 2 This is a flowchart of a dataset construction method proposed in an embodiment of this application, as shown below. Figure 2 As shown, the method includes the following steps:

[0052] Step S201: Obtain samples from multiple groups of metal samples, conduct corresponding tests on the samples for multiple performance parameters to be tested, and obtain the values ​​of multiple performance parameters of the metal samples prepared under each set of process parameters.

[0053] Specifically, in this embodiment, when the method is applied to prepare aerospace titanium alloy Ti-6Al-4V using the LPBF process, the required performance parameters for the titanium alloy sample are tensile strength and elongation. First, a sheet tensile specimen is obtained by wire cutting the prepared Ti-6Al-4V titanium alloy sample.

[0054] Then, tensile tests were conducted using a universal testing machine to obtain tensile strength and elongation data of Ti-6Al-4V titanium alloys prepared with different process parameters.

[0055] Step S202: Summarize the values ​​of each group of process parameters and the corresponding multiple performance parameters to generate an additive manufacturing process parameter and performance parameter data table.

[0056] Specifically, the tensile strength and elongation data of each group of laser powder bed melting process parameters and the corresponding metal samples are summarized. In this embodiment, the mechanical property data of the Ti-6Al-4V alloy obtained from the laser powder bed melting process parameters and the test are summarized in Table 1 below:

[0057] Table 1. Parameters and Mechanical Properties of Ti-6Al-4V Alloy Powder Bed Melting Process

[0058]

[0059]

[0060] As shown in Table 1, this embodiment summarizes the tensile strength and elongation of metal samples prepared under five different sets of process parameters.

[0061] Step S203: Normalize the data table to obtain a dataset reflecting the relationship between process parameters and performance.

[0062] Specifically, the data summarized in step S202 is normalized to scale various data types to a specific range, thereby improving computational efficiency and performance in subsequent data analysis and modeling. Normalization methods that can be used include, but are not limited to, min-max normalization, Z-score normalization, and decimal scaling normalization.

[0063] Thus, after normalization, a dataset of Ti-6Al-4V laser powder bed melting process parameters and mechanical properties was obtained.

[0064] Step S102: Using the dataset as training data, and multiple process parameters as model inputs, a single performance prediction model for predicting individual performance parameters is trained using a Gaussian regression algorithm. The combined performance is obtained by weighting multiple performance parameters, and a combined performance prediction model for predicting the combined performance is trained using a Gaussian regression algorithm.

[0065] Specifically, to address the high cost of multi-objective parameter optimization, this step employs a Gaussian process regression model to model multiple objectives (i.e., performance parameters) separately. Based on the data normalization process in the previous step and the weighted processing in this step, the complex multiple objectives can be transformed into a single objective (i.e., comprehensive performance).

[0066] In practice, model training data can be generated first based on the dataset constructed above. For example, historical data and data collected through other means reflecting the relationship between metal additive manufacturing process parameters and sample performance can be added to the existing dataset. Then, using each process parameter as input and the performance parameter as output, a Gaussian process regression method is used to train a prediction model for process parameters-single performance. Next, using predetermined weighting coefficients, the required performance parameters are weighted to obtain the comprehensive performance, and a Gaussian process regression method is used to train a prediction model for process parameters-comprehensive performance.

[0067] The process of training the prediction model using the Gaussian regression algorithm can refer to the method described in the relevant embodiments. For example, first select the kernel function, then estimate the hyperparameters, then train the model based on the model assumptions, and finally perform prediction and model evaluation. As an example, the Gaussian process regression model uses the radial basis function (RBF) as the kernel function, sets the initial length scale to 1.0, and the length scale range is

[10] . -2 10 -2 The noise variance α is set to 10. -4 In the hyperparameter optimization, the number of optimizer restarts was set to 10, and the output values ​​of the model were the predicted values ​​and standard deviations of tensile strength and elongation, respectively.

[0068] In one embodiment of this application, a weighted average of multiple performance parameters is used to obtain a comprehensive performance, including: setting a weighting coefficient for each performance parameter based on user preferences and the importance of each performance parameter; multiplying each weighting coefficient by the value of the corresponding performance parameter; accumulating the obtained multiplications to calculate a comprehensive performance index; and using the comprehensive performance index as training data for a comprehensive performance prediction model.

[0069] Continuing with the above example, based on the constructed dataset, using laser powder bed melting process parameters as input and tensile strength and elongation properties as output, Gaussian process regression is used to train a prediction model for Ti-6Al-4V laser powder bed melting process parameters-single mechanical properties.

[0070] Then, based on user preferences (which can be determined by combining historical data and current production needs) and the importance of each performance indicator (which can be determined by combining expert knowledge and historical experience), the weighting coefficients for tensile strength and elongation are set to 0.7 and 0.3, respectively. The measured performance values ​​in Table 1 are then weighted to obtain a comprehensive performance index; for example, the weighting coefficients are multiplied by the corresponding measured performance parameters, and then the products are summed. This comprehensive performance index can be used to train the comprehensive performance prediction model. Gaussian process regression is then used to train the process parameter-comprehensive performance prediction model.

[0071] As one possible implementation, the optimization method of this application can be realized by using the intelligent optimization system for metal additive manufacturing process parameters based on Bayesian optimization proposed in this application. Figure 3 It intuitively demonstrates the specific process of using the system to execute the optimization method of this application and the interaction process between various objects.

[0072] Step S103: Set the range of multiple process parameters, use the comprehensive performance prediction model as the surrogate model in Bayesian optimization, select the acquisition function for the additive manufacturing process, and perform Bayesian optimization based on the range and the acquisition function.

[0073] Specifically, a Bayesian optimization framework is first constructed, then the range of metal additive manufacturing process parameters is defined, and then the process parameter-comprehensive performance prediction model obtained in step S102 is used as a surrogate model. A suitable acquisition function is selected, and Bayesian optimization is performed.

[0074] The range of process parameters can be determined based on the current optimized additive manufacturing process, taking into account factors such as the general specifications of the process and the current optimization requirements. By setting the range, the range of target parameters for Bayesian optimization is defined to limit the search space and avoid wasting computational resources in invalid or unreasonable regions. For example, for the example of preparing Ti-6Al-4V alloy by laser powder bed melting process, the range of process parameters is as follows: laser power: (100, 1000), scanning speed: (200, 2000), powder layer thickness: (10, 100), scanning spacing: (20, 200), powder particle size: (20, 100), preheating temperature: (25, 500).

[0075] It should be noted that the comprehensive performance prediction model obtained in this application is used as a surrogate model, which can simulate complex solution processes with low computational cost. In Bayesian optimization, this surrogate model can be used to model the objective function to efficiently find the global optimum. The selected acquisition function is a key tool for guiding the search for the next evaluation point, achieving a balance between exploration and utilization.

[0076] In one embodiment of this application, the selectable acquisition functions include, but are not limited to, Expected Improvement (EI), Probability of Improvement (PI), and Upper Confidence Bound (UCB). When the additive manufacturing process to be optimized is laser powder bed melting, the selected acquisition function is Expected Improvement (EI).

[0077] Step S104: Based on the Bayesian optimization results, determine the recommended process parameters that make the objective function of the comprehensive performance prediction model optimal, and predict the predicted values ​​of each performance parameter of the metal sample under the recommended process parameters using a single performance prediction model.

[0078] Specifically, this step uses a single-objective driven Bayesian optimizer to efficiently recommend the optimal combination of process parameters. Based on the Bayesian optimization results in step S103, the combination of process parameters that optimizes the comprehensive objective function is selected as the recommended process parameters. Then, using the single-performance prediction model trained in step S102, the various properties of the metal sample under the recommended process parameters are predicted, and the predicted values ​​of each performance index (i.e., predicted values) are obtained.

[0079] Continuing with the example above, based on the Bayesian optimization results, the optimal combination of process parameters with the best overall performance is recommended. A pre-trained Gaussian process regression model is then used to predict the tensile strength and elongation corresponding to the recommended process parameters. In this embodiment, when using the above system for optimization, the initial optimization results are as follows: Figure 4 As shown, the recommended optimal process parameter combination is as follows: laser power: 388.6W, scanning speed: 1173.4mm / s, powder thickness: 91.8μm, scanning spacing: 48.7μm, average powder particle size: 53.3μm, and substrate preheating temperature: 271.8℃.

[0080] Furthermore, the system predicts that the tensile strength of the Ti6Al4V titanium alloy prepared under this optimal combination of process parameters is 1251 MPa and the elongation is 9.2%.

[0081] Step S105: Perform performance tests on the additive manufacturing samples prepared under the recommended process parameters. Based on the error between the measured values ​​and the predicted values, update the dataset and iteratively train and optimize the prediction model until the corresponding process parameters that meet the error requirements are obtained.

[0082] Specifically, this step automatically iterates the process parameter optimization model based on the deviation between the measured and predicted performance values ​​under the optimal combination of process parameters, thereby improving the quality of the final recommended process parameters and the reliability of the decision. In practice, metal additive manufacturing samples are prepared under the recommended process parameter conditions, and their performance is tested. The measured performance results are compared with the performance parameter results predicted by the model in step S104, and the error between the prediction and the measured performance is calculated. Then, the process parameter optimization model is automatically iterated based on this error value.

[0083] In one embodiment of this application, based on the error between the measured value and the predicted value, the dataset is updated and the prediction model is cyclically trained and Bayesian optimized, including the following steps:

[0084] Step S1: If the error corresponding to each performance parameter is within the preset allowable range, the current recommended process parameters are taken as the final process parameters.

[0085] Specifically, in this embodiment, the error is the ratio of the absolute value of the difference between the measured value and the predicted value to the predicted value. If the errors corresponding to each performance parameter are all within the preset allowable range, then the currently recommended optimal process parameters are adopted, and optimization is stopped.

[0086] Step S2: If the error corresponding to any performance parameter exceeds the allowable range, add the measured value and the recommended process parameter to the dataset to update the dataset.

[0087] Specifically, if the error corresponding to any performance parameter exceeds the preset allowable range, the experimental data obtained from the new test (including the process parameters recommended through Bayesian optimization, as well as the performance parameter results obtained from prediction and actual measurement) will be added to the dataset constructed in step S101 to obtain an updated dataset.

[0088] Step S3: Retrain the single performance prediction model and the comprehensive performance prediction model using the updated dataset, and perform Bayesian optimization again on the retrained comprehensive performance prediction model.

[0089] Specifically, using the updated dataset, steps S102 to S105 are re-executed to update and train the multi-objective prediction model, and Bayesian optimization is performed again until the optimal process parameters that meet the requirements are obtained, that is, the error corresponding to each performance parameter of the metal sample prepared under the optimal process parameters is within the preset allowable range.

[0090] Continuing with the example above, experimental verification was conducted under the recommended process parameters to prepare Ti6Al4V titanium alloy samples and test their tensile strength and elongation, which were 1225 MPa and 9.8%, respectively. Figure 3 As shown, the measured performance is entered into the system and feedback is submitted. The system automatically calculates the error between the prediction and the measured performance. If the performance error is within the preset allowable range, the currently recommended optimal process parameters are adopted and optimization stops. If the error exceeds the preset allowable range, the experimental data obtained from the new test (including process parameters and performance results) is added to the dataset constructed in step S101, and steps S102 to S105 are re-executed to update and train the multi-objective prediction model. Bayesian optimization is then performed again until the optimal process parameters that meet the requirements are obtained.

[0091] In this example, such as Figure 5 As shown, the errors in tensile strength and elongation reported by the system are 2.1% and 6.5%, respectively, while the preset allowable range is less than or equal to 10%. According to the above description, the error requirement is met. Therefore, the current combination of process parameters is adopted as the optimal process, and the process optimization ends.

[0092] In summary, the Bayesian optimization-based intelligent optimization method for metal additive manufacturing process parameters in this application addresses the problems of high cost, long cycle, and small datasets in acquiring additive manufacturing process parameter-performance data. It employs Gaussian process regression modeling to reduce the number of experiments and the amount of data collected, thereby lowering the cost and time of process parameter optimization and improving optimization efficiency. Furthermore, for multi-objective optimization problems, this method transforms the multi-objective optimization problem into a single-objective optimization problem, efficiently exploring the parameter space through Bayesian optimization to quickly find the optimal trade-off point between multiple conflicting objectives and recommend the optimal combination of process parameters. Moreover, this method possesses a closed-loop feedback mechanism from prediction to verification, automatically iteratively optimizing the process parameter optimization model based on the deviation between measured and predicted values, improving the quality of the final recommended process parameters and the credibility of the decision.

[0093] To achieve the above embodiments, this application also proposes an intelligent optimization system for metal additive manufacturing process parameters based on Bayesian optimization. Figure 6 This is a schematic diagram of the structure of an intelligent optimization system for metal additive manufacturing process parameters based on Bayesian optimization, as proposed in an embodiment of this application. Figure 6 As shown, the system includes: a construction module 100, a training module 200, an optimization module 300, a prediction module 400, and an update module 500.

[0094] The construction module 100 is used to prepare multiple sets of metal samples under multiple process parameters with different values ​​using the additive manufacturing process to be optimized, and to test multiple performance parameters of the multiple sets of metal samples to construct a dataset reflecting the relationship between process parameters and performance.

[0095] The training module 200 is used to take the dataset as training data, take multiple process parameters as model input, train a single performance prediction model for predicting a single performance parameter through Gaussian regression algorithm, and weight multiple performance parameters to obtain comprehensive performance, and train a comprehensive performance prediction model for predicting comprehensive performance through Gaussian regression algorithm.

[0096] The optimization module 300 is used to set the range of multiple process parameters, use the comprehensive performance prediction model as a surrogate model in Bayesian optimization, select the acquisition function for the additive manufacturing process, and perform Bayesian optimization based on the range and the acquisition function.

[0097] The prediction module 400 is used to determine the recommended process parameters that make the objective function of the comprehensive performance prediction model optimal based on the Bayesian optimization results, and to predict the predicted values ​​of each performance parameter of the metal sample under the recommended process parameters through a single performance prediction model.

[0098] The update module 500 is used to perform performance tests on additive manufacturing samples prepared under recommended process parameters. Based on the error between the measured values ​​and the predicted values, the dataset is updated and the prediction model is trained and Bayesian optimization is performed iteratively until the corresponding process parameters that meet the error requirements are obtained.

[0099] As another possible implementation, one embodiment of this application also proposes a Bayesian-based intelligent optimization system for metal additive manufacturing process parameters. This system includes:

[0100] The data entry module is used to verify the rationality of the input additive manufacturing process parameters and performance, and to record them, thereby constructing a dataset of metal additive manufacturing process parameters and performance.

[0101] The model training module is used to select appropriate machine learning algorithms, train performance prediction models based on datasets, and weight target metrics according to actual needs.

[0102] The optimization calculation module is used to construct a Bayesian optimization framework, use a multi-objective prediction model as a surrogate model, execute the Bayesian optimization algorithm, and determine the optimal process parameters.

[0103] The results output module recommends the optimal combination of process parameters, predicts its performance using a multi-objective prediction model, and displays and stores the optimization results.

[0104] The verification feedback module is used to compare and analyze the measured performance results of the sample with the predicted results based on the recommended optimal process parameters, and to determine whether the optimization has been completed or to automatically execute the next iteration of optimization.

[0105] It should be noted that the foregoing explanation of the embodiment of the intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization also applies to the system of this embodiment, and will not be repeated here.

[0106] In summary, the Bayesian optimization-based intelligent optimization system for metal additive manufacturing process parameters in this application reduces the cost and time of process parameter optimization and improves process optimization efficiency. It can quickly find the optimal trade-off point among multiple conflicting objectives and recommend the optimal combination of process parameters. Furthermore, this system improves the quality of the final recommended process parameters and the reliability of the decision.

[0107] To implement the above embodiments, this application also proposes an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the Bayesian-based intelligent optimization method for metal additive manufacturing process parameters as described in any of the first aspects above.

[0108] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization as described in any one of the first aspects of the embodiments above.

[0109] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0110] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0111] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0112] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0113] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0114] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0115] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0116] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for intelligent optimization of metal additive manufacturing process parameters based on Bayesian optimization, characterized in that, Includes the following steps: Using the additive manufacturing process to be optimized, multiple sets of metal samples were prepared under multiple process parameters with different values, and multiple performance parameters of the multiple sets of metal samples were tested to construct a dataset reflecting the relationship between process parameters and performance. The dataset is used as training data, and the multiple process parameters are used as model inputs. A single performance prediction model for predicting individual performance parameters is trained using a Gaussian regression algorithm. The multiple performance parameters are then weighted to obtain the comprehensive performance. A comprehensive performance prediction model for predicting the comprehensive performance is then trained using a Gaussian regression algorithm. Set the range of the multiple process parameters, use the comprehensive performance prediction model as a surrogate model in Bayesian optimization, select a data acquisition function for the additive manufacturing process, and perform Bayesian optimization based on the range and the data acquisition function. Based on the Bayesian optimization results, the recommended process parameters that make the objective function of the comprehensive performance prediction model optimal are determined, and the predicted values ​​of each performance parameter of the metal sample under the recommended process parameters are predicted by the single performance prediction model. The performance of the additive manufacturing samples prepared under the recommended process parameters is measured. Based on the error between the measured value and the predicted value, the dataset is updated and the prediction model is trained and Bayesian optimization is performed iteratively until the corresponding process parameters that meet the error requirements are obtained.

2. The method according to claim 1, characterized in that, The additive manufacturing process includes: laser powder bed melting, electron beam powder bed melting, and direct energy deposition. The multiple performance parameters include: density, hardness, yield strength, tensile strength, elongation, surface roughness, and residual stress.

3. The method according to claim 2, characterized in that, When the additive manufacturing process to be optimized is laser powder bed melting, the multiple process parameters include: laser power, scanning speed, powder bed thickness, scanning spacing, powder particle size, and substrate preheating temperature. The dataset reflecting the relationship between process parameters and performance includes: Samples are obtained from the multiple groups of metal samples, and corresponding tests are performed on the samples for multiple performance parameters to be tested, so as to obtain the values ​​of multiple performance parameters of the metal samples prepared under each set of process parameters. Summarize the values ​​of each set of process parameters and the corresponding multiple performance parameters to generate an additive manufacturing process parameter and performance parameter data table; The data table is normalized to obtain the dataset reflecting the relationship between process parameters and performance.

4. The method according to claim 3, characterized in that, The weighted processing of the multiple performance parameters to obtain the comprehensive performance includes: Based on user preferences and the importance of each performance parameter, a weighting coefficient is set for each performance parameter; Each weighting coefficient is multiplied by the corresponding performance parameter value, and the resulting multiplications are accumulated to calculate a comprehensive performance index. This comprehensive performance index is then used as training data for the comprehensive performance prediction model.

5. The method according to claim 2, characterized in that, The acquisition function includes: expected improvement, probabilistic improvement, and upper confidence limit. When the additive manufacturing process to be optimized is laser powder bed melting, the acquisition function selected is the expected improvement.

6. The method according to claim 1, characterized in that, The process of updating the dataset based on the error between the measured values ​​and the predicted values, and iteratively training and optimizing the prediction model, includes: If the error corresponding to each performance parameter is within the preset allowable range, the current recommended process parameters will be used as the final process parameters. If the error corresponding to any performance parameter exceeds the allowable range, the measured value and the recommended process parameter are added to the dataset to update the dataset; The single performance prediction model and the comprehensive performance prediction model are retrained using the updated dataset, and Bayesian optimization is performed again on the retrained comprehensive performance prediction model.

7. The method according to claim 6, characterized in that, The error is the ratio of the absolute value of the difference between the measured value and the predicted value to the predicted value. The allowable range is less than or equal to 10%.

8. A smart optimization system for metal additive manufacturing process parameters based on Bayesian optimization, characterized in that, Includes the following modules: The module is used to prepare multiple sets of metal samples under multiple process parameters with different values ​​using the additive manufacturing process to be optimized, and to test multiple performance parameters of the multiple sets of metal samples to construct a dataset reflecting the relationship between process parameters and performance. The training module is used to use the dataset as training data, the multiple process parameters as model inputs, train a single performance prediction model for predicting a single performance parameter using a Gaussian regression algorithm, and perform weighted processing on the multiple performance parameters to obtain the comprehensive performance, and train a comprehensive performance prediction model for predicting the comprehensive performance using a Gaussian regression algorithm. The optimization module is used to set the range of the multiple process parameters, use the comprehensive performance prediction model as a surrogate model in Bayesian optimization, select a data acquisition function for the additive manufacturing process, and perform Bayesian optimization based on the range and the data acquisition function. The prediction module is used to determine the recommended process parameters that make the objective function of the comprehensive performance prediction model optimal based on the Bayesian optimization results, and to predict the predicted values ​​of each performance parameter of the metal sample under the recommended process parameters through the single performance prediction model. The update module is used to perform performance tests on the additive manufacturing samples prepared under the recommended process parameters, update the dataset based on the error between the measured values ​​and the predicted values, and iteratively train and optimize the prediction model until the corresponding process parameters that meet the error requirements are obtained.

9. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent optimization method for metal additive manufacturing process parameters based on Bayesian optimization as described in any one of claims 1-7.

Images